Research And Application On Ascites Cancer Cells Of Image Processing And Recognition

It is a medical problem how to quickly and accurately diagnose cancer which has plagued doctors. In many cases of the diagnosis cancer, because lesions of the benign and the malignant tumor are difficult to be distinguished, and some lack of experienced doctors can not diagnose the disease in time, the best opportunity of treatment for patients is delayed. In particular, there are more and more features of cancer cell types, so it will be more and more difficult to diagnose the cancer. However, with the development of modern computer technology the digital computer technology is used in many industries, people see the dawn using the computer of characteristics of rapid and accurate to diagnose the cancer.Cancer cells image recognition technology is to identify the cancer the main purpose of cancer cells image recognition, which is a simulation of the process of cancer diagnosis observing cells image under the microscope using a digital computer, extracting the cell image feature inner the complex background using image processing technique, converting these image features into numerical form and combining with the experience and knowledge of doctors, is to identify the cancer.So far, there are many theoretical approaches of Image Recognition at home and abroad, but effective recognition system is not formed. There are some major problems satisfying what is difficult to distinguish and split adhesion cells, to precisely extract the nuclear characteristics, and is that the recognition accuracy is not high. In this paper, in order to solve the problem of cell adhesion, the distance transform algorithms is used , which is by modeling watershed theory to effectively separate cell adhesion; spiral mode of growth is used in the region growing to quickly and accurately split the nucleus and cytoplasm and to extract the feature of the nucleus and cytoplasm as the basis for cancer cell image recognition; there are a variety of cell characteristics ,the standard to determine cells is imprecise , so fuzzy pattern recognition method is very useful to effectively distinguish diseased cells and normal cells. Experimental results show the image processing method is very effective ,which is used to solve problems of adhesion cell division and nuclear extract; fuzzy pattern recognition method has a higher recognition rate in terms of identifying diseased cells than the others .In the test, cancer cell image recognition system in this article identify the 150 image samples and recognition rate retches about 65% and much higher than others , it has a high recognition efficiency, the main reason is that fuzzy identification methods do not need too many samples to identify and get diseased cells.